Artificial Intelligence in Industrial Markets

ARTIFICIAL INTELLIGENCE IN INDUSTRIAL MARKETS

, the Compass icon and the 3DS logo, CATIA, SOLIDWORKS, ENOVIA, DELMIA, SIMULIA, GEOVIA, EXALEAD, 3D VIA, 3DSWYM, BIOVIA, NETVIBES, and 3DEXCITE are commercial trademarks trademarks commercial are 3D VIA, 3DSWYM, BIOVIA, NETVIBES, and 3DEXCITE EXALEAD, SOLIDWORKS, ENOVIA, DELMIA, SIMULIA, GEOVIA, CATIA, and the 3DS logo, icon , the Compass EXPERIENCE® 3D Our 3DEXPERIENCE® platform powers our brand applications, serving 12 industries, and provides a rich portfolio of industry solution experiences. Dassault Systèmes, the 3DEXPERIENCE® Company, provides business and people with virtual universes to imagine sustainable innovations. Its world-leading solutions transform the way products are designed, produced, and supported. Dassault Systèmes’ collaborative solutions foster social innovation, expanding possibilities for the virtual world to improve the real world. The group brings value to over 230,000 customers of all sizes in all industries in more than 140 countries. For more information, visit www.3ds.com. ©2014 Dassault Systèmes. All rights reserved. Systèmes. ©2014 Dassault approval. written their express is subject to or its subsidiaries Systèmes trademarks owners. Use of any Dassault owned by their respective are All other trademarks other countries. or its subsidiaries in the U.S. Systèmes and/or of Dassault trademarks or registered

Americas Europe/Middle East/Africa Asia-Pacific Dassault Systèmes Dassault Systèmes Dassault Systèmes K.K. 175 Wyman Street 10, rue Marcel Dassault ThinkPark Tower Waltham, Massachusetts CS 40501 2-1-1 Osaki, Shinagawa-ku, 02451-1223 78946 Vélizy-Villacoublay Cedex Tokyo 141-6020 USA France Japan WHY ARTIFICIAL INTELLIGENCE MATTERS TO INDUSTRIAL MARKETS Industries like manufacturing, mining and construction are sometimes characterized as digital Luddites, with an odd attachment to business relics like fax machines, paper catalogs, clipboards and Post-it Notes. US$690B In reality, of course, industrial companies have been at the forefront of technological innovation Approximate amount IDC for decades, with sophisticated robots and other complex, highly-automated machinery widely forecasts discrete and process deployed on the shop floor and in the field. manufacturing will spend on digital transformation solutions Nonetheless, it is true that there is a real, widely-acknowledged gap between this operational in 2023, representing nearly technology and business-centered information technology (IT). Industrial markets have long 30% of total global trailed others in their level of IT investment. This, however, is about to change. DX spending. Industrial companies are now expected to outspend their B2C counterparts on digital IDC Worldwide Semiannual transformation solutions. Will this investment yield the same kind of disruption and Digital Transformation transformation seen in industries like retail, banking, insurance and health care? This remains Spending Guide to be seen. Digital Transformation is More Than Simple Digitalization Global market intelligence firm IDC has found that most manufacturers’ digital transformation investments will not yield the results they seek. Why? For some, it will be because their efforts are not really geared toward “digital transformation” but rather “digitalization”, which is focused primarily on improving efficiency. While efficiencies are desirable (who doesn’t want to save time, lower costs, or reduce waste?), the changes resulting from digitalization are not transformational, and the yields are likely to be minimal: industrial companies have already spent years using dozens of strategies to squeeze out every ounce of production and supply chain efficiency they can. Digital transformation, on the other hand, goes far beyond efficiency. It is a strategy that encompasses digitalization, yes; but it also enables continuous and substantive process improvement, increased agility and, most importantly, breakthrough innovation. What is Digital Transformation? IT analyst firm Gartner, Inc., defines “digital business transformation” as “the process of exploiting digital technologies and supporting capabilities to create a robust new digital business model.” It’s an important definition in that it places “business” literally in the middle of “digital transformation,” and it foregrounds the development of a new business model that is data-centered (“a new digital business model”). It is the kind of definition Jeffrey Immelt, former Chairman and CEO of General Electric, had in mind as he mused on the impact of the “If you went to bed last night as an industrial Internet of Things (IoT) on industrial companies. His advice to his company, you’re going to wake up today as a peers? Digital transformation is going to reshape your market more software and analytics company.” profoundly and more rapidly than you can imagine: “If you went Jeffrey Immelt, former Chairman & CEO to bed last night as an industrial company, you’re going to wake up General Electric today as a software and analytics company.” Specifically, digital transformation involves using data, analytics, and connectivity to rethink everything you do from the perspective of your customer, and using these same tools as the backbone of new products and services for a compelling, personalized customer experience. Let‘s see how this kind of deep transformation can be achieved via advanced technologies like artificial intelligence and machine learning.

2 Artificial Intelligence in Industrial Markets INDUSTRIAL APPLICATIONS OF ARTIFICIAL INTELLIGENCE & MACHINE LEARNING The use cases for artificial intelligence (AI) and machine learning (ML) in industrial markets (see Appendix) are quite broad and diverse, but behaviorally they are somewhat similar. Like a valued butler, a good AI or ML application is a helpmate, anticipating needs, managing tasks, and providing trusted advice (recommendations). Below are some examples of the kind of valuable assistance AI and ML can provide throughout a product’s or an asset’s lifecycle. Predictive Maintenance and Field Operations The most commonly cited industrial application of AI is predictive maintenance, e.g. the ability to predict when an equipment failure will happen to avoid expensive downtime costs. Few people realize how much AI can be used upstream, downstream, and beyond these specific AI-powered predictive models in order to design and scale the benefits of AI to all field operations. ML-enabled field support can assist with: • Maintenance, Repair and Operations planning; • Generation of preventive and predictive maintenance recommendations; • Analysis of quality issues; • Automation of routine operations and maintenance tasks using automation software, robots, autonomous vehicles and drones; • Interpreting and funneling operational data back to teams working on service design; and • Interpreting and sharing performance data, along with other quality-related data like customer feedback and warranty information, to manufacturing teams.

BENEFITS OF PREVENTIVE MAINTENANCE

Improved reliability and life of equipment

Fewer costly repairs and downtimes associated with unexpected equipment failure

Fewer errors in operations as a result of equipment working incorrectly

Reduced health and safety risks

Design In the conceptual phase of product development, ML is applied in combination with virtual engineering models and simulation for iterative design. With these technologies, millions of design options can be cycled through in an instant, with recommendations automatically generated for optimal solutions based on multiple criteria (cost, sustainability, time, regulatory requirements, etc.). AI and ML are also valuable in the earliest ideation stages. They are being used in cognitive search systems to help designers explore existing design concepts via both text and image searches. And they can help designers understand customer demand through analysis of sources like social media or internal customer feedback systems.

Artificial Intelligence in Industrial Markets 3 Testing ML can also be used to develop highly accurate digital models of both physical objects and systems. This enables the development of realistic behavioral models that can be used to run performance simulations. This use case is so well-established that physical prototyping has been all but eliminated in some industries (architecture, automotive, aerospace). Manufacturing ML-powered digital modeling and simulation (including virtual reality systems) are also being used to 1) plan production lines and systems, 2) develop and integrate smart equipment, smart robots and production-line drones, 3) recommend and execute proactive maintenance (preventive/predictive maintenance), and 4) funnel important production data back to teams working on product design and specifications.

Factory Flow Simulation

Sales & Marketing In commercialization phases, AI and ML applications are being used to: • Predict demand trends, • Deliver highly personalized/micro-targeted marketing, • Create intelligent, multi-lingual bot assistants for self-service ordering and support, • Power sales- and marketing-related virtual and augmented reality applications, and • Customize products and services.

Drones are being integrated into logistics systems A 3D plant model can be used for training

4 Artificial Intelligence in Industrial Markets Summary Benefits of Machine Learning in Industrial Contexts The most significant payoffs of using ML in industrial sectors include greater innovation, process optimization and industrialization, automation, and improved quality, with the latter often yielding the quickest returns. Real-time monitoring systems for early detection of issues, advanced cognitive systems for issue investigation, predictive maintenance systems, and feedback loops for design teams have proven to be helpful aids in enhancing quality and reducing direct and indirect costs related to rework, waste, warranty claims, and recalls. When solutions like these and other ML applications share common digital models of products and assets, and are deployed via a common collaborative platform, they can also provide the consistent digital thread (digital continuity) needed for continuous product and process innovation. This makes immediate wins in quality improvement an important lever for exponential gains through accelerated innovation.

KEY CHALLENGES OF ARTIFICIAL INTELLIGENCE IN INDUSTRIAL SECTORS To reap the full value of quality improvements and other potential AI and ML benefits, several important challenges will need to be overcome. Scaling AI and ML to the organization requires more than just the right amount of data and the right infrastructure. Scaling ML requires collaboration. Collaboration vs. Data Silos - In-Context Information is Still Stuck in Silos One of the core challenges of ML is providing it with the proper fuel, that is to say the right data. Getting the right data usually requires tackling enterprise silos, including data silos and silos of analysts and analytical products. On the data side, this means combining disparate types of internal and external data (sound, images, text, 3D files, structured data, etc.) in order to provide the full context for design, production or field operations issues. This context is valuable not only because it enables smarter, better-informed analyses and recommendations, but because it gives human beings greater confidence in relying on machine-generated answers.

ANALYTICS CHALLENGE: INFORMATION SILOS

WWW Design Quality DB

PLM Email

CRM Social Media Open Data

IoT Maintenance Supply Chain System SCADA Partner Systems

Artificial Intelligence in Industrial Markets 5 Collaboration for Trust in Data - Black Box/White Box Balance This issue of confidence is important in ML for business. Executives and operations teams often demand to understand the recommendations and algorithms used to produce results, but this too often forces data scientists to sacrifice model efficacy in order to boost legibility. A better option is to allow specialists to craft optimal algorithms and to share their work and data with peers, and to ”whitebox” only the context for non-specialists, providing visibility and drill down into the parameters and data used, but not the algorithms themselves. Collaboration to Scale the Benefits to the Broader Organization Likewise, it is important to industrialize ML outputs that prove to be of value. This means packaging and integrating them in a platform to make them accessible to all enterprise applications, across all teams and lifecycle processes. Using a platform strategy to break down analytical silos also enables critical governance for ML products, supporting standardization, certification, IP protection, personnel training (a perennial challenge), and traceability. Without such governance and industrialization, ML cannot truly support continuous process or product innovation and improvement. But with the right discipline and technology choices, these challenges are all addressable. What’s most important now is to define the right digital business platform for your business in order to enable the right collaboration capabilities across your organization, and set your company on the path to leveraging AI and ML for your digital transformation. An evaluation should be performed to identify the internal capabilities and talents required in each of these collaboration areas, and a blueprint developed for leveraging these internal and external resources. With competition tight for data science professionals, a more efficient collaboration will be mandatory.

REALIZING THE VALUE OF ARTIFICIAL INTELLIGENCE IN INDUSTRIAL SECTORS Digital Continuity & Digital Twins Digital continuity is about creating an environment where all information from every phase of a product or asset’s lifecycle - from “Digital continuity is the core characteristic that concept to disposal or reuse - is captured in real time and transformed we need for the 21st century digital world.” into actionable insights. Michael Grieves, Executive Director, Center for Manufacturers have long wanted this kind of full lifecycle visibility, Advanced Manufacturing and Innovative Design but they have been frustrated by either siloed data, or data deserts Florida Institute of Technology (nonexistent data for certain objects and events). Now, enhanced data access technologies and sensor-based IoT connectivity can deliver the comprehensive data and data views needed. What’s more, a “digital twin” of a physical process, product, asset or environment can provide a unique, authoritative and consistent referential for this lifecycle data. For instance, a 3D CAD model of a product can provide a referential to which data can be linked from conception through to design, engineering, manufacturing and post-sales service.

Going a step further, a simulation-capable twin, like Dassault Systèmes’ virtual twin experience, provides more than just a consistent, “single-version-of-the-truth” referential. It is an extremely powerful tool for iterating through scenarios and options for everything from design to fabrication to maintenance and repair, for continuous improvement and innovation. And the more real-world data the digital model is fed, the more accurate and valuable are the simulations it supports. The Right Digital Collaboration Infrastructure Digital transformation requires a general foundation of social, mobile, cloud and Internet technologies. Connecting people, places, and objects, and enabling anywhere, anytime collaboration, is at the heart of the digital age. For digital continuity and the proper functioning of digital twins, a business platform with integrated search capabilities is a requirement. This type of platform can connect to or crawl all relevant internal and external resources, and provide unified views of both structured and unstructured information across these silos. With the right platform, these views can include dashboard metrics and recommendations based on advanced analytics.

6 Artificial Intelligence in Industrial Markets Artificial Intelligence in Industrial Markets 6 And if that platform is 3D-based like the 3DEXPERIENCE platform, decision makers can view and drill down on KPIs that are integrated into 3D renderings of products, assets, or environments. This provides an intuitive visual means of reviewing performance for all stakeholders, as well as providing designers with immediate, in-context feedback on the performance impacts of their design choices. As with recommendations, these performance simulations are powered by advanced analytics. Advanced Analytics for Informed Decisions in Context Traditional business analytics are descriptive in nature. They provide summary or detailed views of a current state of affairs (e.g., project progress or spending reports) or an analysis of Advanced analytics seek to historical events (e.g., lifecycle cost or revenue for a product line). They are “what is” and “what answer “what will be” or has been” analytics. “what should be” questions Advanced analytics go further. They enrich such descriptions with critical context, helping though predictive and decision makers to not only better understand “what is,” but to make more accurate “what will prescriptive analytics. be” predictions, and wiser “what should be” decisions. Advanced analytics can be performed using traditional mathematical or statistical techniques, or newer, big data-enabled ML strategies. It is these newer strategies which are key to digital transformation in industrial markets. Industrial companies already have masses of under-utilized structured and unstructured data scattered across active and legacy systems, and the Industrial IoT (IIoT) is adding enormous new streams of valuable real-time data. And according to McKinsey & Co., the IIoT is especially valuable during challenging times (see below). To sift through and make sense of this big data, transforming it into business insights and new products and services, discipline is required. Organizations need to enrich it with context and enable digital continuity. This will facilitate more efficient collaboration and, as a result, it will be much easier to develop AI strategies in general, and ML techniques in particular.

Artificial Intelligence in Industrial Markets 7 7 Artificial Intelligence in Industrial Markets ARTIFICIAL INTELLIGENCE & MACHINE LEARNING AT DASSAULT SYSTEMES The NETVIBES brand reveals Information Intelligence on the Dassault Systèmes 3DEXPERIENCE platform. It enables organizations to gather, align and enrich big data - whether internal or external, structured or unstructured, simple or complex, real-time or archived - and to deliver that information to users within high-value applications. The brand’s portfolio of solutions has incorporated big data management technologies and advanced analytics, including ML tools and techniques, since its inception. Specifically, its first product was a Web-scale search platform designed for enterprise use. It used ML to enrich data by surfacing hidden information and relationships, and successfully integrated advanced analytics into its sub-second query processing framework. As such, NETVIBES provided an ideal foundation for the development of ML technologies on the 3DEXPERIENCE platform, and its big data processing framework remains the core engine powering the brand’s suite of enterprise applications. Sourcing & Standardization Intelligence Sourcing & Standardization Intelligence solutions address four common and costly problems about how to: define product part standards and share them across projects; choose whether to reuse an existing part from a previous project, buy it from a supplier, or create a new one; equip purchasing departments with the right level of knowledge to improve cost efficiency; and drive reuse in engineering in accordance with the company’s sourcing and standardization policy. These problems are solved by AI and ML techniques that automatically group parts together based on their shape similarity and semantic characteristics, as well as aggregating technological and business information to break down the silos between engineering and procurement. The result is a pre-identified classification of parts that complexity and sourcing ONE-THIRD Amount of time engineers managers can review and use to launch appropriate actions to either de-duplicate the part report they waste on non- references or renegotiate with suppliers. value-added work In addition, the PartSupply service on the 3DEXPERIENCE Marketplace makes comparative How to Reduce Non-Value- data available on tens of millions of catalogued parts from more than 1,000 suppliers. Users Added Work in Engineering, can search to evaluate technical characteristics, performance and quality, and quickly secure Tech-Clarity vendors in their area able to deliver.

8 Artificial Intelligence in Industrial Markets Artificial Intelligence in Industrial Markets 8 THE HIGH COST OF A NEW PART The U.S. DoD estimates the average lifecycle cost of a new part for a weapons system is $27,500

ACTIVITY COST ($) % OF TOTAL CUMULATIVE COST COST ($) Engineering 12,600 45% 12,600 & Design Testing 1,000 4% 13,600

Purchasing 5,200 19% 18,800

Manufacturing 2,400 9% 21,200

Inventory 1,200 4% 22,400

Logistics Support 5,100 19% 27,500

TOTAL 27,500 - -

Source: SD-19 Parts Management GuideU.S. Dept. of Defense

PLM Analytics Aimed at bridging the gap between virtual and real worlds, the PLM Analytics portfolio allows customers to fully leverage advanced analytics around product data along with related projects, issues, changes and quality processes. Uncovering such metrics is especially challenging because related data is heterogeneous, ranging from technical notes, 2D drawings and 3D models to multiple bills of materials stored in various formats. To overcome this challenge, NETVIBES uses specific search and aggregation mechanisms and ML algorithms to produce mathematical representations of these diverse objects, allowing them to be integrated in relevant context for analytics purposes. One solution using this strategy is Asset Quality Intelligence (AQI). Dedicated to improving the quality and reliability of assets in their operation phase, the AQI cognitive search engine uses internal and/or external data to detect anomalies and emerging trends, while automatically classifying and clustering multi-source field issues. The use of ML in this context provides a full 360° view of similar issues. It also enables valuable discovery with deep learning techniques used (when data volumes are sufficient) to surface and qualify issues. The solution is also designed to satisfy the need for transparency with sufficient visibility into the data used to aid comprehension, confidence, and certification - while avoiding over-simplifying and weakening algorithms - in order to expose them in a way that is comprehensible to non-specialists (for example, all the parameters and filters used during indexation are exposed, even if complete indexation algorithms are not). Additional solutions also apply ML techniques to solve custom analytics challenges. Developed with customers, they cover many use cases around 360° fleet views, 360° customer views, analysis of customer churn risk, sales prospect scoring, customer targeting, discovery of adverse effects in FDA data and manufacturing procedures, regulation discovery and matching for individual specifications or change requests, and new supplier discovery. Social Business NETVIBES Social Business solutions equip enterprises to listen, learn, and act on all the social Web data that matters to their business. ML-enabled algorithms allow users to analyze these data and extract relevant insights on all-in-one dashboards for informed decision-making.

Available on the Dassault Systèmes 3DEXPERIENCE platform, Information Intelligence solutions support continuous product improvement and integration, and enable essential analytics governance and traceability (history/versioning), ultimately helping customers improve innovation, operational excellence and business performance.

Artificial Intelligence in Industrial Markets 9 9 Artificial Intelligence in Industrial Markets APPENDIX: DEFINITIONS

Artificial Intelligence Artificial intelligence (AI) is conceptual in nature. It is a term that is used in a general way to describe the capacity of machines to imitate some aspect of human intelligence. Accordingly, AI is a label often applied to software programs that perform cognitive tasks that humans can also do, like filtering junk email, interpreting handwriting, detecting fraud, translating speech, and recommending products. But these types of applications can perform their functions with or without the aid of advanced analytics and behavioral models. They can execute their functions with or without any capacity to “learn,” that is to say, to independently improve their performance over time. Certainly, however, their performance can be greatly enhanced with such attributes. Moreover, many would argue that the use of models and the capacity to “learn” are defining characteristics of “true” AI.

COMMON USES OF AI

Robotics Natural Interfaces Cognitive Systems • Factory Automation • Speech Recognition • Smart Agents/Bots • Exoskeletons • Natural Language • Expert Systems • Autonomous Vehicles • Eye & Hand Gestures • Recommendation • Security Systems • Brain-Computer Engines • Environmental Interfaces • Adaptive Learning Monitoring • Touch Surfaces Systems • Remote Exploration • Virtual Reality • Translation Systems

Regardless of the level of precision one wishes to attach to AI, however, machine learning is inarguably a core - if not the core - technology used to implement AI strategies today for all types of tasks. But what, precisely, is machine learning? Machine Learning Machine learning (ML) is a body of techniques that enables computers to learn how to execute tasks without having every step and every possible option scripted via software programs (i.e., without relying on rules-based programming). This is accomplished by processing data using ML An algorithm is simply a algorithms. discrete, step-by-step set of procedures, like a recipe, An algorithm is simply a discrete, step-by-step set of procedures, like a recipe, that enables a that enables a computer to computer to solve a specific problem or perform a particular task. While algorithms are used for solve a specific problem or both traditional software programming and ML, the “task” a ML algorithm performs is to build a perform a particular task. (reusable) model that describes or predicts something about the world. Inherent in the “learning” part of the definition is the idea that the models will improve over time with exposure to new data or feedback about the accuracy of predictions or descriptions. A baseline ML model can be developed using training sets of real or simulated data prepared under the supervision of a human being (hence it is called “supervised learning”). For example, in the case of specific types of equipment malfunctions, a scientist may feeda computer a set of carefully annotated documents detailing past malfunctions. These documents are used to train the computer to identify patterns associated with various types of malfunctions, thereby enabling it to detect possible future instances of these malfunctions.

10 Artificial Intelligence in Industrial Markets Alternatively, a scientist may use simulation software to generate a very large number of simulated malfunction scenarios (alone or in tandem with real, historic data) to help train a malfunction-detection algorithm. In unsupervised learning, the emphasis shifts from training a computer with the types of real or synthetic examples, to equipping it to sift through mountains of data and detect meaningful signals on its own (for example, identifying clusters of performance anomalies for a piece of equipment without being trained to look for specific malfunctions). Unsupervised learning works best with very large data sets, and therefore computer-generated virtual data – which can be produced in massive quantities – is particularly useful. Virtual data is also valuable for a particular type of ML framework known as deep learning, which can be unsupervised, supervised, or semi-supervised, but always requires an extremely large data set. Deep Learning Deep learning is a ML framework patterned after the neural networks of the brain. It involves processing extremely large amounts of data in steps, or layers, with the results of each layer passed to the next layer to cumulatively build a complex and (hopefully) accurate digital model of a real-world behavior, like playing a game of chess or driving an autonomous car. 125B Capgemini estimates “that by 2020, 125 billion devices could be connected through IoT.” With Estimated number of devices the rise of the IoT/IIoT, millions of new sensors are being deployed daily to help gather the big connected through IoT by 2030 data that deep learning thrives on, like periodic or streaming measures of pressure, volume, Capgemini temperature, direction, etc. This sensor data - especially when combined with virtual data and with other types of real data like images, audio files and text documents - can fuel high-value AI applications.

Artificial Intelligence in Industrial Markets 11 LEARN HOW TO PUT ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING commercial are 3DEXCITE and IFWE NETVIBES, BIOVIA, 3DSWYM, VIA, 3D EXALEAD, GEOVIA, SIMULIA, DELMIA, ENOVIA, SOLIDWORKS, CATIA, logo, 3DS the icon, Compass the , TO WORK FOR YOUR COMPANY AT: www.3ds.com/netvibes EXPERIENCE® 3D Our 3DEXPERIENCE® platform powers our brand applications, serving 11 industries, and provides a rich portfolio of industry solution experiences. Dassault Systèmes, the 3DEXPERIENCE Company, is a catalyst for human progress. We provide business and people with collaborative virtual environments to imagine sustainable innovations. By creating ‘virtual experience twins’ of the real world with our 3DEXPERIENCE platform and applications, our customers push the boundaries of innovation, learning and production. Dassault Systèmes’ 20,000 employees are bringing value to more than 290,000 customers of all sizes, in all industries, in more than 140 countries. For more information, visit www.3ds.com trademarks or registered trademarks of Dassault Systèmes, a French “société européenne” (Versailles Commercial Register # B 322 306 440), or its subsidiaries in the United States and/or other countries.All other trademarks are owned by their their by owned are trademarks other countries.All other and/or reserved. States rights United All the Systèmes. in Dassault subsidiaries ©2021 its or 440), 306 322 B # Register Commercial (Versailles européenne” “société French a Systèmes, Dassault of trademarks registered or trademarks approval. written express their to subject is trademarks subsidiaries its or Systèmes Dassault any of Use owners. respective